LLMs are capable, at best, of very limited forms of reasoning. The absence of abstract latents precludes multiple inferences and the optimized action to perform

LLMs are capable, at best, of very limited forms of reasoning. The absence of abstract latents precludes multiple inferences and the optimized action to perform a task.